Photographer Catherine Hyland (CNN)
A few days ago, I've come across an article on a Chinese news website, talking about ghost cities in China had been identified by Baidu Big Data Lab. The ghost city phenomenon has long been noticed and reported by news agencies, but it always lacked the concrete statistic evidence behind those horrified pictures like above. Out of curiosity, I searched on Google for the specific paper and below is a brief summary of this study.
Baidu’s Big Data Lab uses big data analytic to measure housing vacancy rate and locate where these ghost cities lie in real time. The data source is from Baidu's own user data (include temporal and spatial information) and residential area pinpointed at Baidu Map. The data collection lasted over 6 month in 2014 and 2015. After initial data cleaning and processing, Baidu used clustering algorithm to analyze users' spatial distribution and then calculate the vacancy rate based on Baidu Map data.
In order to solve the problem of classifying ghost cities, the team used two different standard to calculate the ranges for normally populated residential areas. The final result has considered both measures mentioned. Below is a picture pinpointed ghost city areas in 9 cities by Baidu.
Vacant areas in 9 cities (Baidu Big Data Lab)
Baidu has offered the list of cities that are detected the existence of vacant residential areas and label the cities that are heavily rely on tourism. Since none of our team member comes from these place, it's hard for us to decide whether the result from Baidu actually reflects the real situation. However, based on the my personal experience, there's something Baidu has missed.
This study of "ghost cities" has revealed a corner of what role Big Data Analytic can play in urban planning. In addition to real estate, railways, airports and roads are also sometimes suffer from messy urban planning. There are roads that has few vehicles on them, while there are roads that packed with vehicles. Roads have been constantly mended since the traffic exceeds expectation. Is it possible to use Big Data Analytic to monitor the road condition? Such urban planning probably can be improved with the help of Big Data.
For developing countries with big population like China and India, building infrastructure is hard. Of course, money, technology and labor are needed, but how to avoid the appearance of "ghost city" remains a challenges. According to macroeconomic principles, market itself sometimes need intervention and I think "ghost city" phenomenon is an example of that. People tend to think of solution when the problem appears, but Big Data has provided real-time data and evidence for officials to make decisions at the planning stage.
Right now, Baidu's strategy on Big Data has mainly focused on business use. The study mentioned above is part of the big data analytic package designed for real-estate clients. However, this study has further proved the over-construction problem in urban China, and I personally hope more Big Data studies will emerge and offer a new perspective on some traditionally challenging problems.
Reference:
[1] Data mining reveals the extent of China ghost cities http://www.technologyreview.com/view/543121/data-mining-reveals-the-extent-of-chinas-ghost-cities/
[2] "Ghost Cities" analysis based on positioning data in China http://arxiv.org/abs/1510.08505
[3] China's New ghost town: Wonderland in Beijing http://travel.cnn.com/shanghai/life/chinas-new-ghost-town-nankou-wonderland-beijing-561846
- Multiple real estates owned by one family: There are majorly two reasons for an apartment to be vacant, one is that someone bought it but didn't actually move in, another is that this place has never been sold or finished. To exclude the first type of situation, Baidu has already tried to exclude the new residential projects. Even though it is hard to get the sales progress of an open real estate project, it is a common situation in China which should not be completely overlooked. I think if Baidu could include the data from secondary housing market and rental market, the result can be more robust.
- Predict the next "ghost city": There are a lot of indicators that calculates the tendency of becoming the next"ghost city" on news reports, but they mostly are based on the change of local housing unit price and recent land auction price. Utilizing Baidu's real-time analytic, the "ghost city" can be identified in a more timely manner.
This study of "ghost cities" has revealed a corner of what role Big Data Analytic can play in urban planning. In addition to real estate, railways, airports and roads are also sometimes suffer from messy urban planning. There are roads that has few vehicles on them, while there are roads that packed with vehicles. Roads have been constantly mended since the traffic exceeds expectation. Is it possible to use Big Data Analytic to monitor the road condition? Such urban planning probably can be improved with the help of Big Data.
For developing countries with big population like China and India, building infrastructure is hard. Of course, money, technology and labor are needed, but how to avoid the appearance of "ghost city" remains a challenges. According to macroeconomic principles, market itself sometimes need intervention and I think "ghost city" phenomenon is an example of that. People tend to think of solution when the problem appears, but Big Data has provided real-time data and evidence for officials to make decisions at the planning stage.
Right now, Baidu's strategy on Big Data has mainly focused on business use. The study mentioned above is part of the big data analytic package designed for real-estate clients. However, this study has further proved the over-construction problem in urban China, and I personally hope more Big Data studies will emerge and offer a new perspective on some traditionally challenging problems.
Reference:
[1] Data mining reveals the extent of China ghost cities http://www.technologyreview.com/view/543121/data-mining-reveals-the-extent-of-chinas-ghost-cities/
[2] "Ghost Cities" analysis based on positioning data in China http://arxiv.org/abs/1510.08505
[3] China's New ghost town: Wonderland in Beijing http://travel.cnn.com/shanghai/life/chinas-new-ghost-town-nankou-wonderland-beijing-561846
